Raman Spectroscopy

    Introduction

The meaning and the importance of an artwork do not reside in the matter that constitutes it but in the meaning it expresses. However, the durability of an artistic work inevitably depends on the materials of which it is composed. Knowledge of these materials is indispensable, not only for the purpose of restoration and conservation, but also for a better and deeper appreciation of the artwork. 

Restoring and conserving a work of art should be preceeded by a thorough study and characterization of all the constituent materials. In this respect, heritage science comes close to material science, and the evolution of techniques for characterization of materials has enabled the evolution of our knowledge of artworks. This is the case of most of spectroscopic techniques, in particular of non-destructive techniques such as Raman spectroscopy. 

Characterization of materials includes two aspects: knowledge of chemical elements or chemical composition and their spatial distribution or chemical structure. Spectroscopic techniques such as XRF and LIBS give information about chemical composition; IR and Raman spectroscopy are able to give information about the chemical structure. These techniques have been useful in dating and authentication, in provenance and technology studies and especially in restoration and conservation. 

Recently, very exhaustive and interesting reviews and books, such as Raman Spectroscopy in Archaeology and Art History, (H.J. Edwards and J.M. Chalmers (Eds.), 2005, Royal Society of Chemistry, UK) have been published, where reader is given a comprehensive review of the progress of the application of Raman spectroscopy in art [1-5]. 

In the analysis of archaeological objects and artworks, Raman spectroscopy has important advantages over other techniques:
-    analyses are non-destructive, not only regarding the sample but also regarding the object, because no sampling and no sample preparation is needed. The analysis causes neither damage nor alteration. The analysis can be do in-situ on large and non-uniformly shaped objects.
-    It is possible to analyse a very small area (1 μm) because laser can be focussed on a very small spot. If it is possible, the amount of sample needed to perform an accurate analysis is very small (some μg). With confocal and imaging configurations it is also possible to obtain stratigraphic information; providing 3D maps of the chemical distribution of all the constituent species of the sample.
-    With the help of optical fibres, laser beam can be delivered remotely and the scattered light can be delivered back to a detector. Light, compact and portable Raman spectrometers can be used [6].
-    The analysis is very rapid and the results are very easy to manage; comparison of a Raman spectrum of the sample with a library of spectra is possible and an unambiguous match leads to compound identification [7-11].

However, Raman microscopy is not suitable for all types of materials. For example, some organic compounds give a very intensive fluorescence spectrum that prevents assignment of Raman bands. Quantitative data are also difficult due to their dependence on individual instrumental parameters.

The study of art objects by Raman spectroscopy is an area of research in rapid evolution. Since 1993, when the first reports were published, the increasing numbers of papers during the last years reflect the attractiveness of this technique, as shown in Fig. 4.1.1.

In 1997, the first special issue of Journal of Raman Spectroscopy on applications of Raman in art was published. Also in 2004, a special issue of this journal, corresponding to the “2^nd International Conference on the Application of Raman Spectroscopy to Art and Archaeology” (Ghent University, 2003) was published. In August 2005 the “3^rd International Conference on the Application of Raman Spectroscopy in Art and Archaeology” took place at the University Pierre et Marie Curie in Paris.

4.1.2          Principles

Raman spectroscopy involves molecular and crystal lattice vibrations and is therefore sensitive to the composition, bonding, chemical environment, phase, and crystalline structure of the sample material in any physical form: gases, liquids, solutions, and crystalline or amorphous solids. 

The Raman phenomenon is a consequence of sample illumination with a monochromatic photon beam (laser), most of which are absorbed, reflected, or transmitted by the sample. However, a small fraction of photons interacts with the sample. Duirng this interaction, some energy is transmitted to elementary particles of which materials are constituted (electrons, ions etc.). This causes their transition from ground energy levels to ‘virtual’ excited states. These excited states are highly unstable and particles decay instantaneously to the ground state by one of the following three different processes:
-    Rayleigh scattering: the emission of a photon of the same energy allows the molecule to relax to its ground vibrational state (elastic scattering). Rayleigh scattering, therefore, bears no information on vibrational energy levels of the sample.
-    Stokes and anti-Stokes Raman photons (inelastic scattering): emission of a photon with an energy either below or above that of Rayleigh photons, thereby generating a set of frequency-shifted ‘Raman’ photons. The energy differences of the Stokes and anti-Stokes Raman photons with respect to the excitation energy give information about molecular vibrational levels.

These potons are collected by a detector and transformed to electrical signals and finally to the corresponding Raman spectrum. Usually, Stokes bands which are more intense than anti-stokes bands are called “Raman spectrum” of the sample (Fig. 4.1.2). The Rayleigh band is filtered out before the detector. 

Raman spectroscopy, discovery by C.V. Raman in the 1920s, remained without interest for decades due to the lack of an efficient monochromatic source and appropriate detectors. Raman spectroscopy was born again with the invention and availability of lasers and CCD detectors.

More detailed treatises can be found in “Raman Microscopy, developments and applications” (G. Turrell & J. Corset, Eds., Academic Press, 1996). Reports on the most recent applications of Raman spectroscopy can be found in “Modern Raman Spectroscopy–A Practical Approach” (E. Smith & G. Dent, Eds., John Wiley & Sons, Chichester, 2005).

4.1.3       Raman Spectrometers

A typical Raman spectrometer is made up of five basic parts (Fig. 4.1.3):
-    Excitation source (generally a laser): A laser is used to produce Raman spectra because it gives a coherent beam of monochromatic light. This gives sufficient intensity to produce a useful amount of Raman scatter.
-    Sample illumination and scattered light collection system (probe): The probe is a collection device that collects the scattered photons, filters out the Rayleigh scatter and any background signal from the fibre optic cables, and sends the scattered light to the spectrograph. Many probes also focus and deliver the incident laser beam.
-    Sample holder
-    Spectrograph: When Raman-scattered photons enter the spectrograph, they are passed through a transmission grating to separate them by wavelength and are passed to a detector.
-    Detection system (optical multichannel analyser, PMT, intensified photo array, or a charged coupled device, CCD): this records the intensity of the Raman signal at each wavelength. This data is represented as a Raman spectrum.

The monochromatic light from a laser passes through focussing optics and a beamsplitter to the sample. The scattered light passes through the beamsplitter to a detector. 

The laser light illuminates the sample through a microscope objective (magnification from 100x to 1000x, typically), which is used both for the illumination (laser beam coming from the laser through mirrors and/or optical fibre) and collection of the scattered light. The scattered light goes to the CCD detector via two steps, the first one to suppress the Rayleigh scattering and the second one to split the selected spectral window on the CCD array in order to be able to see all the spectral components. A computer is used to scan, collect, and process the data creating a graph showing the intensity of light at each wavelength. The change in energy is observed as a change in frequency of the incident beam upon scattering.

The microscope offers visualization of the studied area and of the laser spot, which facilitates the choice of the appropriate region to be analyzed. 

A large variety of laser beams are in use, UV (250 nm), visible (green, 514 nm, or red, 633 nm), near-IR (780 nm or 1064 nm). The intensity of the Raman signal is proportional to 1/λ⁴, where λ is the laser wavelength. As a consequence, UV lasers produce Raman bands of a higher intensity and IR lasers produce the least intense Raman spectra. Unfortunately, selection of the ideal laser is not easy because UV lasers are expensive and spectral resolution is not optimal. On the other hand, IR lasers produce a weak Raman signal. In the middle, visible lasers give a good signal/background ratio but in some cases fluorescence spectra of a sample overlap with the Raman spectrum which prevents assignation of Raman bands. In general, green laser (514 nm) leads to more intensive fluorescence than red laser (780 nm), as shown in Fig. 4.1.4.

In general, three types of Raman spectrometer configuration are possible.

4.1.3.1       Large-Size, High-Resolution Instruments

This type of multichannel spectrometer (Fig. 4.1.5) is equipped with a double monochromator as a filter and a liquid nitrogen-cooled CCD matrix (2000 x 800 pixels) detector. Resolution and accuracy are high (<0.5 cm^-1) and the Raman spectrum can be recorded down to ~10 cm^-1. The required power of illumination on samples is typically between 0.1 and 10 mW. The corresponding recording times range between a few tens of seconds to a few hours, depending on sample composition. Laser illumination and collection of scattered light can proceed through lenses (macroconfiguration) or through a (confocal) microscope (magnification up to 1000x, resolution ~1 μm).

4.1.3.2       Compact Notch-Filtered Instruments

This type of multi-channel instrument is equipped with a pre-filter stage, which consists of "Notch" filters set according to the laser in use. Air- and Peltier effect-cooled CCD detectors are used. Sensitivity is maximal so that the required power of illumination can be decreased to a few μW, which facilitates the examination of black samples and collection of images. Typical recording times per point range between a few seconds and tens of minutes. It is not possible to record Raman spectra below 120-150 cm^-1 (unless very expensive, short-life notch filters are used). Wavenumber resolution and accuracy are limited to ~2-3 cm^-1. Note that this type of instrument can be coupled to horizontal and vertical microscopes or with an optic fibre connected to a remote optic head (fibre length of ~5-10 m, see Fig. 4.1.3). 

4.1.3.3       Portable Instruments

The basic parts of a portable instrument (Fig. 4.1.6) are similar to those discussed in the preceeding section. However, sample manipulation parts are absent, and the analyzed spectral range is defined by the fixed grating in combination with the laser wavelength. Consequently, the grating is optimized for the wavelength and sensitivity is maximal. Transportation and mounting are easy, and the equipment is ready for use in ~10 min. The only required facility is an electric plug.

All spectrometers must be calibrated before analysis. There are two common methods: calibration against a spectral discharge lamp; and calibration against a reference material, such as Si, whose Raman spectrum has been carefully measured. Periodic recalibration is recommended even if there have been no changes in probe or slit position.

4.1.4       Information Obtained from Spectra

Raman signal is collected by a detector and a computer creates a graph showing the intensity of light at each wavenumber. The change in energy is observed in terms of "Raman shift" with respect to excitation frequency of the incident beam, while the magnitude of the shift itself is independent of the excitation frequency. This Raman shift is therefore an intrinsic property of the sample. In general, only some excitations of a given sample are "Raman active," that is, only some may take part in the Raman scattering process. The peaks in the intensity occur at the frequencies of the Raman active modes. For a transition to be Raman active there must be a change in polarizability of the molecule. 

There are selection rules that govern the ability of a molecule to be detected using Raman spectroscopy. Each molecular bond is characterised by unique energy transitions and subsequent shifts in wavelength of the original beam (Raman shift). Measuring the wavelength shift allows the identification of molecular species on the sample surface. Therefore, Raman spectroscopy provides details on the chemical composition, molecular structure, and molecular interactions. 

Raman spectra provide "fingerprints" (Fig. 4.1.7) of the molecular structure and, as such, permit qualitative analysis of individual compounds, either by direct comparison of the spectra of the known and unknown materials run consecutively, or by comparison of the spectrum of the unknown compound with catalogues of reference spectra. By comparisons with the spectra of a considerable number of compounds of known structure, it may be possible to recognize bands at specific positions in the spectrum, which can be identified as "characteristic group frequencies" associated with the presence of a particular molecular structure, such as methyl, carbonyl, or hydroxyl groups. 

Raman spectrometry is a method of determining modes of molecular motion, especially the vibrations, and their use in analysis is based on the specificity of these vibrations. It is predominantly applicable to qualitative and quantitative analysis of covalently bonded molecules rather than ionic structures. Real-time chemical analysis can be performed in a non-contact manner.

The reader should be reminded, though, that many interesting samples are fluorescent; and each sample will require multiple acquisitions at short exposure times. It is feasible to remove backgrounds (fluorescence, baseline and noises) from spectra, however, integration times must be short to avoid saturation of the detector and it may be necessary to average several spectra to obtain an adequately large signal to noise ratio. Some experimentation is usually required to determine the optimum conditions.

Fluorescence is often observed in artworks materials. Such fluorescence extends over thousands of cm^-1and often arises from organic materials on the surface of an artwork. Cleaning with water, ethanol or other solvents is not recommended; it is possible to clean by exposure to a powerful violet or blue laser in ~20 min.

4.1.5          Case Studies

One of the first applications of Raman spectroscopy in art was the identification of pigments in different supports [12]. Nowadays, this application is also the most frequent. Several papers report results on identification of organic and inorganic pigments found in different solid materials: paper [13] (manuscripts, cantorals, bibles, lithographs, papyri, stamps and wallpapers), easel paintings, walls (frescoes [14] and rock art [15]), ceramic and Egyptian artefacts [16]. 

In some cases, specific important groups of pigments such as blue, white and red pigments [17] have been studied by Raman spectroscopy. 

Unambiguous identification of compounds using Raman spectroscopy was possible also to many artworks, where no sampling is possible, such as archaeological textiles, wood, ivory and jade art objects [18]. 

Studies of corrosion processes on metals are also possible [19]. Such knowledge may help in dating and provenance studies, and in studies of cleaning and restoration of metallic artefacts. Similar applications have been published on patinas, where binders, adhesives and resins had been used. 

Usually, the study of glasses, ceramics and glazes involves only determination of chemical composition because application of classical techniques for structural determination gives no information on disordered materials. Raman spectroscopy is a good tool to resolve this problem [20]. In the last years, comparisons of structures of different ceramics, glazes and glasses have helped to determine their provenance. 

Raman spectroscopy is often applied in forensic analysis, and in the same manner, studies on archaeological human remains and bio-organisms have been successfully carried out [21].

Below, we report on some representative case studies.

4.1.5.1          Pigment Analysis of the Gutenberg Bibles

The printing of 180 copies of the Bible was the major achievement of Johann Gutenberg. The three years project was completed in 1455 in Gutenberg’s workshop in Mainz, Germany. The surviving copies and fragments are held at academic institutions and libraries throughout the world or in private collections. Some of these copies were also illuminated, depending on the geographical location, taste, and wealth of the owner. 

It is therefore of interest to determine the palettes used in the illuminated copies of Gutenberg bibles, to establish whether the palettes differ from one part of Europe to another, whether Europe in effect constituted a fairly coherent cultural unit in terms of pigment use in book art or whether local preferences and resources are reflected. 

Raman analyses of seven copies of the Gutenberg bible located in Europe have been carried out. The pigment palette has been established as cinnabar/vermillion, azurite, calcite (chalk), basic lead carbonate (lead white), lead tin yellow type 1, lead(II,IV) oxide, malachite, an organo-copper complex (verdigris), carbon-based black, gold leaf, an organic red pigment, red and yellow ochres, indigo, and lazurite (Fig. 4.1.8). Some of these components are present only in more precious copies. The presence of anatase and rutile in one copy may be due to recent restoration work or contamination from other sources. In general, the palettes have many similarities even though the styles of illumination vary considerably.

The application of non destructive method of analyzing the pigments used for illuminations has been successful in allowing the determination of palettes of those books to which access is limited by time, equipment availability, etc. The results provide an understanding of the practical aspects of the illuminator’s work.

4.1.5.2          Characterization of Mummified Tissues

In this work, the authors examined whether NIR-FT-Raman spectroscopy could be employed to determine to what extent human remains have been preserved, which have been subjected to artificial mummification. In addition, the purpose was to shed light on the methods and complex mummification techniques used by ancient Egyptians. Raman microscopic studies of human skin from the XIIth dynasty mummy, Nekht-Ankh, ca. 2100 BC, from the ‘Tomb of the Two Brothers’ (Khnum-Nakht and Nekht-Ankh) from Rifeh in Middle Egypt (Fig. 4.1.9), excavated from a hot desert environment, was carried out. The mummies were embalmed by natron, which absorbs water and is also mildly antiseptic, in the way used by ancient Egyptians.

The amide I and III protein Raman bands were used to monitor the presence of proteins and to give an idea about the secondary protein structure. The lipids and proteins seemed well preserved, although different degrees of protein deterioration were observed. In some spots, protein degradation was extensive. Some sites showed very well preserved protein secondary structures with both helical and sheet contents, associated protein modes at 1660, 1450, and 1240 cm^-1 (Fig. 4.1.10) indicating that the artificial mummification process had a positive effect although no embalming chemicals were left in those spots. Of significant interest is that the presence of residual chemicals in the skin tissues (Na₂SO₄ from the natron desiccant used in the mummification process), is correlated with the presence of the most degraded organic tissue and hence the most poorly preserved skin structure (Figs. 4.1.11, 4.1.10 and 4.1.11 with permission of J. Raman Spectrosc. 34 (2003) 375). 

The conclusion is that the artificial embalming process used by the ancient Egyptians was an efficient way to preserve the mummies even under hot conditions.

4.1.5.3          Ancient Ceramics and Glasses

Raman spectroscopy allows non-destructive remote analysis of crystalline and amorphous phases as ceramics, glaze and nanosized pigments in glasses. In this work, selected glasses and glazes of various porcelains, celadons, faiences and potteries, representative of different production technologies used in the ancient, European, Mediterranean, Islamic and Asian cultures were studied. Their identification is based on the study of Raman fingerprints of crystalline and glassy phases. Raman parameters allow for the classification as a function of composition and/or processing temperature. 

Ceramics are artificial rocks obtained by firing mixed raw materials together, which are more or less transformed by the thermal treatment. Ancient ceramics are composites of sintered grains of different compositions. Raw materials are almost fully molten to produce a glass or a glaze. On the other hand, unreacted, incompletely dissolved raw materials as well as some phases formed during the process, are crystalline. Different kinds of products are obtained if different technologies are applied to the same starting batch or if a given technology is applied to raw materials processed differently. Moreover, different technologies will often give products of very similar appearances (“blue” porcelains, Fig. 4.1.12), although these products are completely different in their micro/nanostructure. A lot of information on the process remains written in the sample and non-destructive Raman analysis of the microstructure (for ceramics) and nanostructure (for glasses and enamels) offers a way to identify it and, sometimes, to date ancient artefacts.

From these Raman spectra, the identification of phases and polymorphs is possible through comparison with spectral databases or reference materials analyzed simultaneously by X-ray diffraction. Examples are given by the comparison between typical 18^th-century hard- and soft-paste porcelains or between an original Vietnamese celadon and its copy (Fig. 4.1.12).

4.1.5.4          In-Situ Analysis of Medieval Frescoes with Portable Raman Spectroscopy

The non-destructive nature of Raman spectroscopy makes this method suitable for in-situ analysis of artefacts. The results and difficulties experienced during analysis of wall paintings on the ceiling of a mediaeval chapel of Ponthoz (Belgium) using the mobile Raman spectrometer MARTA, are especially interesting (Fig. 4.1.13).

Although the mediaeval wall paintings from the chapel of the castle of Ponthoz are well-preserved, some interesting degradation phenomena could be observed: the identification of a black degradation product, likely to be meta-cinnabar, a degradation product of the red pigment vermilion (HgS); the formation of gypsum (CaSO₄•2H₂O) as a weathering product of calcium carbonate (CaCO₃); the observation of copper(II) hydroxychlorides.

As the frescoes are located on the ceiling of the chapel, some practical problems had to be solved. These problems were related to the instrumental set-up, positioning of the probe head and influence of the local environment during analysis. To overcome the two first issues, the equipment was brought onto wooden platform constructed at a height of ca. 4 m above ground level and remote-controlled motorized micro-translation stages were used to avoid vibrations caused by the operator. In order to overcome local environmental problems (low temperature and direct light entering through stained glass windows), portable heating and black curtains had to be used.

Despite the unfavourable conditions, several pigments were identified using in-situ Raman spectroscopy. Most of the Raman spectra showed the presence of calcite, applied as the supporting layer. Raman spectroscopic inspection revealed that the blue areas contain azurite but in some degraded area this colour was changed to green due to the formation of clinoatacamite.

4.1.6          Conclusions

The number and diversity of the applications cited in this Chapter prove the potential role that Raman microscopy can play in artwork characterisation and conservation. The specific and non-destructive structural identification of materials can provide the art historians, archaeologists and conservators invaluable information regarding authenticity, provenience, manufacturing technology, trade patterns, state of preservation and even approximate age of artworks. The advantages of molecular spectroscopic identification of organic and inorganic materials along with elemental identification have proved to be of value to analysts working at the interface of art history, museum science, chemistry and biology. 

In conclusion: why should Raman spectroscopy be considered in studies of artworks? Because the technique is:
-    non-destructive and non-intrusive,
-    no sample preparation is required,
-    the Raman phenomenon depends on molecular structure and physical form; identification of chemical species and bonding interactions are possible,
-    With microscopy, spatial resolution of ~1 μm, depth of field ~1 μm can be achieved,
-    Spectra are generally well resolved and with high information content,
-    Samples can be solids, liquids and gases, transparent or opaque,
-    In situ real-time and in-air measurement,
-    Laser wavelength strongly influences the fluorescence but changing the laser wavelength displaces fluorescence out of the spectral region of interest,
-    Identification of phases,
-    Identification of crystalline polymorphs,
-    Measurement of mid-range order in solids,
-    Phase transition and order-disorder transitions in minerals (phase transition in quartz, graphite